epoch_forge/docs/issue-audit.md

4.9 KiB

Gitea issue audit

This audit maps the open issues from the project tracker to the current codebase as of 2026-09-22. Tracker entries predate several milestones, so an open checkbox does not necessarily mean the behavior is absent.

Issue Current assessment Next action
#19 Tweak Show Addressed: normal rows now use a dedicated short renderer with name, kind, stable ID, position, and aggregate stats. Verify visual density in real terminals.
#18 Implement show Addressed for the CLI: show defaults to all, dispatches every collection, filters by name/text/position, reads historical epochs, and renders epoch summaries at three detail levels. Consider structured output and richer stat predicates later.
#17 Serialization errors/upgrades Version detection and actionable rejection are complete. No legacy release exists, so a speculative upgrader was intentionally removed. Add explicit 1.x migrations when the first real format change exists.
#16 Rewrite backend Addressed by Workspace, typed snapshots, transactional writes, broad integration coverage, and removal of the remaining unjustified legacy unsafe fn declarations. Avoid rewriting stable code wholesale; continue adding focused tests.
#15 Clamp stats Addressed: CLI/TUI mutations clamp normalized persistent stats to 0..=120, event/blocker modifiers and growth to -120..=120, and population remains an unbounded non-negative count (or a bounded negative modifier). Workspace validation rejects bypassed invalid state. Move these defaults into schemas if custom-stat domains become configurable.
#14 Build fsck Addressed for read-only validation: every epoch is parsed and checked for duplicate IDs, collection types, ordinary links, governors, and capitals. Put serializer rewriting in a separately previewed repair command, not fsck.
#13 Add all cards into pre-events Explicitly deferred by the maintainer for this milestone. Revisit only when requested.
#12 Add fight/war Not implemented. The implementation-ready design uses generic multi-side conflict records and reuses deterministic influx resolution. Deliver after the influx engine and diffusion integration; see resolution-conflict-design.md.
#11 Epoch Diffing Addressed: epoch and layer comparisons share a deterministic, serializable typed diff with changed-field details and importance classification. Extend the type rather than parsing terminal output when structured formats arrive.
#10 Display functions Partially addressed: unified show, entity detail, and diff rendering are buffered. Most commands and several legacy library helpers still print directly; saving/structured output is absent. Implement the renderer/sink migration and acceptance inventory in presentation-design.md.
#9 Adding influx Foundation implemented: versioned embedded/override rules, all six levels, deterministic target-isolated rolls, weighted pressure/resistance, four outcomes, attenuation, and serializable typed reports. Persistence and CLI remain open. Persist idempotent resolution records, then implement preview/apply/show/clear.
#8 Event diffusion Partially addressed: deterministic road/direct town frontiers, affected sets, expiry, previewed ticks, and reversible application exist. Influx-aware outcomes, protected-town avoidance, and character targeting remain. Integrate the shared resolution engine using the explicit #8 acceptance gates in resolution-conflict-design.md.
#6 HashMap/HashSet Addressed without format churn: EntityIndex provides ephemeral UUID/name hash indexes and duplicate-ID detection over the stable vector representation. Repository validation uses the index and hash lookup for relationships. Rebuild indexes after mutation; never persist them.

Implementation order is #9's shared resolution engine, the remaining #8 outcomes, then #12 conflicts. #10 presentation work accompanies each slice and finishes with a complete direct-output migration. Issue #13 remains deferred.